Broadcom Inc. is in discussions with lenders to secure more than $60 billion in debt financing for an artificial intelligence chip facility supporting Anthropic and other AI companies. The move reflects the sector's capital intensity: BofA analyst Tom Curcuruto estimates Broadcom's dedicated financing vehicle could accumulate $370 billion in senior debt by mid-2029, with $150 billion in new issuance projected for 2027 alone.
Curcuruto's projections assume a 20-gigawatt operational scale for the underlying AI infrastructure. This architecture—a dedicated debt vehicle financing chip production and deployment—represents a structural shift in how AI capex is being funded. Rather than housing these liabilities on Broadcom's balance sheet, the financing vehicle isolates debt from the parent company's credit profile.
The leverage implications are material. A $370 billion debt facility would dwarf Broadcom's current balance sheet and introduce substantial financial risk if utilization rates or AI capex demand trails projections. Broadcom stock fell 6 percent Friday following Curcuruto's analysis, signaling investor concern about both leverage and execution risk.
The strategy reflects a hard reality: traditional corporate balance sheets cannot absorb AI infrastructure capex at scale. Anthropic, OpenAI, and other frontier labs are driving demand for specialized chips and data center buildout. By creating a dedicated financing vehicle, Broadcom can deploy capital rapidly without constraining its ability to fund core semiconductor operations or return capital to shareholders.
Key risks: If AI capex demand softens or utilization rates fall below Curcuruto's 20-gigawatt assumption, the financing vehicle faces refinancing pressure. Broadcom's capacity to service $370 billion in debt depends on sustained demand from Anthropic and competing AI labs. A slowdown in large language model capex cycles would directly impair the underlying economics.
